JOURNAL OF BONE AND MINERAL RESEARCH
Volume 15, Number 12, 2000
© 2000 American Society for Bone and Mineral Research
Effect of Dietary Protein on Bone Loss in Elderly Men and
Women: The Framingham Osteoporosis Study*
MARIAN T. HANNAN,1,2 KATHERINE L. TUCKER,3 BESS DAWSON-HUGHES,3
L. ADRIENNE CUPPLES,2 DAVID T. FELSON,2,4 and DOUGLAS P. KIEL1
ABSTRACT
Few studies have evaluated protein intake and bone loss in elders. Excess protein may be associated with
negative calcium balance, whereas low protein intake has been associated with fracture. We examined the
relation between baseline dietary protein and subsequent 4-year change in bone mineral density (BMD) for
391 women and 224 men from the population-based Framingham Osteoporosis Study. BMD (g/cm2) was
assessed in 1988 –1989 and in 1992–1993 at the femur, spine, and radius. Usual dietary protein intake was
determined using a semiquantitative food frequency questionnaire (FFQ) and expressed as percent of energy
from protein intake. BMD loss over 4 years was regressed on percent protein intake, simultaneously adjusting
for other baseline factors: age, weight, height, weight change, total energy intake, smoking, alcohol intake,
caffeine, physical activity, calcium intake, and, for women, current estrogen use. Effects of animal protein on
bone loss also were examined. Mean age at baseline (ⴞSD) of 615 participants was 75 years (ⴞ4.4; range,
68 –91 years). Mean protein intake was 68 g/day (ⴞ24.0; range, 14 –175 g/day), and mean percent of energy
from protein was 16% (ⴞ3.4; range, 7–30%). Proportional protein intakes were similar for men and women.
Lower protein intake was significantly related to bone loss at femoral and spine sites (p < 0.04) with effects
similar to 10 lb of weight. Persons in the lowest quartile of protein intake showed the greatest bone loss.
Similar to the overall protein effect, lower percent animal protein also was significantly related to bone loss
at femoral and spine BMD sites (all p < 0.01) but not the radial shaft (p ⴝ 0.23). Even after controlling for
known confounders including weight loss, women and men with relatively lower protein intake had increased
bone loss, suggesting that protein intake is important in maintaining bone or minimizing bone loss in elderly
persons. Further, higher intake of animal protein does not appear to affect the skeleton adversely in this
elderly population. (J Bone Miner Res 2000;15:2504 –2512)
Key words:
protein intake, bone density, elderly, osteoporosis, longitudinal study
INTRODUCTION
STEOPOROSIS IS an important public health problem,
affecting 20 –25 million Americans.(1–3) Although major risk factors have been identified, dietary factors, specif-
O
*Presented in part as concurrent sessions at the 19th and 20th
Annual Scientific Meetings of the American Society of Bone and
Mineral Research, Cincinnati, Ohio, U.S.A. (1997) and San Francisco, California, U.S.A. (1998), as well as the Association of
Rheumatology Health Professionals at the 33rd Annual Scientific
Meeting, Washington, D.C., U.S.A. (1997).
ically macronutrients, represent an important understudied
area in osteoporosis research.(4) Few studies have evaluated
diet and bone loss in free-living elders. Protein is of particular interest and concern because biochemical and nutritional studies from as early as 1920,(5,6) and again more
recently,(7–10) have shown that high protein intake is a
powerful determinant of urinary calcium loss, which could
potentially upset calcium balance and lead to bone loss.
Despite the fact that several studies suggest a role for
protein in bone health, none has examined the effect of
levels of dietary protein on bone loss.
1
Hebrew Rehabilitation Center for Aged, Research and Training Institute and Harvard Medical School Division on Aging, Boston,
Massachusetts, U.S.A.
2
Boston University School of Public Health, Boston, Massachusetts, U.S.A.
3
Jean Mayer United States Department of Agriculture Human Nutrition Research Center on Aging at Tufts University, Boston,
Massachusetts, U.S.A.
4
Boston University Arthritis Center, Boston, Massachusetts, U.S.A.
2504
DIETARY PROTEIN AND BONE LOSS
Dietary protein can cause an increased acid load, which
may be buffered by bone calcium. It is thought that the high
sulfur content of meat may determine an endogenous acid
load that contributes to bone loss. Sebastian et al. showed
that bone loss in 18 postmenopausal women resulted from
mobilization of skeletal calcium salts used to balance acid
from dietary protein metabolism, and that increased acidity
stimulated bone resorption and inhibited bone formation
through suppression of osteoblast function.(11) The increase
in urinary calcium may depend on the amount and type of
protein intake because dietary acid load may be somewhat
influenced by the ratio of meat to vegetable protein intake,(4)
although the effect of type of protein intake on bone remains
unclear.(12–14)
Evidence also suggests that protein undernutrition is associated with osteoporosis. Low protein intake or insufficiency, perhaps a marker of total dietary insufficiency, has
been associated with frailty and fracture in the elderly.(15–17)
Evidence from a nursing home study of fractures and protein intake, although not controlling for confounders, suggested that protein undernutrition may be associated with
osteoporotic fracture.(18) Thus, extreme levels of dietary
protein intake may pose a problem in elderly persons.
To our knowledge, no previous studies have addressed
the possible contributions of dietary protein to bone loss or
considered the effect of animal protein specifically in
population-based groups. Elderly persons are of particular
interest because they often suffer from chronic diseases and
may have a range of protein intakes below amounts previously linked to negative calcium balance. Thus, the purpose
of our study was to examine the association between dietary
protein and the skeletal health of elderly men and women.
We examined the relation between baseline dietary protein
and subsequent 4-year change in femoral, radial shaft, and
spine bone mineral density (BMD) for the elderly members
of the population-based Framingham Study. Further, our
study evaluated the effect of animal protein and nonanimal
protein intake for a possible effect on bone loss in older
individuals.
MATERIALS AND METHODS
Study subjects
The population-based Framingham Cohort was established in 1948 to examine risk factors for heart disease in
5209 men and women of ages 28 – 62 years.(19 –21) Subjects
are seen biennially for a physical examination and a battery
of questionnaires and tests. Since the inception of the primarily white cohort 50 years ago, nearly two-thirds of the
5209 cohort members have died. The surviving, now elderly, cohort subjects follow the same age- and sex-specific
population proportions found in the general Framingham
population.(22) At biennial examination 20 (1988 –1989),
855 cohort members participated in the Framingham Osteoporosis Study in which they had femoral BMD measurements and proximal radial scans and completed a semiquantitative Food Frequency Questionnaire (FFQ). Because of
the length of the routine biennial Framingham Study clinic
2505
examination, subjects were asked to return for a callback
examination to obtain BMDs at the lumbar spine. Results
from the baseline Framingham Osteoporosis Study have
been reported previously.(22)
As part of their regular clinic visit 4 years later (1992–
1993), 615 subjects (72%) who had baseline BMD and the
FFQ assessed had repeat BMD measures. All longitudinal
scan pairs were evaluated for consistency in anatomic site
and quality of analysis by the original technician, and those
scans showing inconsistencies were reanalyzed by two experienced investigators (M.T.H. and D.P.K.). Details on the
longitudinal follow-up osteoporosis examination have been
reported.(23) In brief, the mean 4-year BMD losses were
much greater for women than the losses for men. Annualized rates of mean percent BMD loss are as follows: for
women, femoral neck loss was 0.87%, trochanter was
0.86%, Ward’s area was 1.06%, radial shaft was 1.21%, and
spine was 1.12%; for men, femoral neck loss was 0.38%,
trochanter was 0.04%, Ward’s area was 0.16%, radial shaft
was 0.90%, and spine was 0.09%.
Our study was approved by the Boston University Institutional Review Board, and written informed consent was
obtained for all study subjects at both examinations.
Bone mineral density
BMDs of the proximal right femur (femoral neck, greater
trochanter, and Ward’s area) as well as the lumbar spine
(average BMD of L2–L4) were measured in grams per
square centimeter, using a Lunar dual photon absorptiometer (DP3) at baseline and a dual X-ray absorptiometry
(DPX-L) densitometer (Lunar Radiation Corporation, Madison WI, U.S.A.) at the 4-year follow-up exam. We have
previously shown high correlations between dual photon
and dual X-ray absorptiometry.(24) However, because of a
small but consistent shift in BMD values between the two
technologies, femoral BMDs were adjusted for the change
in equipment from DP3 to DPX-L technology, using published corrections.(24) Bone density at the proximal radial
shaft site was measured in grams per square centimeter
using a Lunar SP2 single photon absorptiometer (Lunar
Radiation Corporation) at both examinations. The right side
was scanned at each exam unless there was a history of
previous fracture or hip joint replacement. For these individuals, the left side was scanned. We used standard positioning as recommended by the manufacturer. Monthly
measurements of a bone phantom over the follow-up period
showed no machine drift across time. At the baseline examination, the coefficients of variation in normals measured
twice with repositioning for the DP3 were 2.6% (femoral
neck), 2.8% (trochanter), 4.1% (Wards area), 2% (radial
shaft), and 2.2% (lumbar spine). At examination 22, the
coefficients of variation for the DPX-L measurements were
1.7% (femoral neck), 2.5% (trochanter), 4.1% (Ward’s
area), and 0.9% (lumbar spine). We examined percent
change in BMD from baseline to 4-year follow-up. Percent
change in BMD was calculated as the difference between
exam 20 BMD and exam 22 BMD, divided by exam 20
BMD, and multiplied by 100.
2506
Food Frequency Questionnaire
Dietary intake was assessed using the 126-item Willett
FFQ(25,26) at the baseline examination in 1988 –1989 and
these data were converted to food and nutrient intake data.
Individuals who reported intakes of less than 600 or greater
than 4000 kcal or with data missing for more than 12 food
items were excluded. The Willett FFQ has been validated,
and nutrient intakes have been shown to correlate well with
those obtained by multiple food records and with blood
measures for several nutrients.(26,27) However, it is only
semiquantitative in nature, and translation of results to
quantitative recommendations remains to be validated for a
number of nutrients. Nevertheless, the Willett FFQ has been
shown to rank individuals well in relation to their actual
intake. In general, usual intake patterns examined for the
purpose of ranking individuals for association with an outcome measure, typically are best measured with an FFQ.
The Willett FFQ has been validated extensively and has
been shown to be reproducible in the long term for both
cohorts of men(25) and women,(26) showing that the FFQ is
a stable and reproducible measure of ranking for food
intakes. As Willett points out,(28) the underlying principle of
an FFQ is that average long-term diet intake is the conceptually important exposure rather than short-term intake
across a few specific days. FFQ methodology therefore
estimates usual intakes over the past year indicating typical
long-term intakes rather than providing precise measurements of short-term intake. To evaluate whether protein
intakes may change over time, we examined the change in
rank from the FFQ measured at baseline in our study to
another measurement of FFQ 2 years later. There were only
minor changes reported in intakes, such that no one in our
study changed quartile of protein intake.
Protein intake
Dietary protein intake was expressed as percent of energy
from protein (percent protein). Further, components of protein were divided into animal and nonanimal protein intake
and expressed as percent of energy from animal protein and
percent from nonanimal protein.
Potential confounding variables: We controlled for the
influence of the following factors at baseline on the relation
between dietary protein intake and bone loss at each skeletal
site: total energy intake, age, sex, weight, height, smoking,
caffeine, alcohol use, physical activity, calcium intake, and
for women current estrogen use. We also included a wellknown risk factor for bone loss: weight change over the
4-year follow-up. Weight in pounds (converted to kilograms) was measured at baseline using a standardized balance beam scale and weight at follow-up was measured
similarly using the same calibrated scale. In addition to
baseline weight, we included percent weight change over
the follow-up as a continuous variable. Height (without
shoes) was measured in inches using a stadiometer and
recorded to the nearest 1⁄4 inch. A participant’s smoking
status was assessed via questionnaire at baseline and preceding examinations as current cigarette smoker at baseline
(smoked regularly in the past year), former smoker, or never
smoked.
HANNAN ET AL.
Caffeine use, incorporating coffee and tea intake at baseline, was defined as the sum of daily coffee (1 cup equals 1
caffeine unit) and daily tea (1 cup equals 0.5 caffeine units)
and grouped into 0 –2 caffeine units consumed per day or
over 2 U/day.(29,30) Weekly use of beer, wine, or hard liquor
at baseline was grouped into ounces of alcohol consumed
per week based on Framingham calculations of intake.(31,32)
Physical activity at baseline was defined using the Framingham Physical Activity Index, a weighted 24-h score of
typical daily activity, based on hours spent doing heavy,
moderate, light, or sedentary activity as well as sleeping.(33,34) Baseline dietary calcium intake, including calcium supplements, was assessed from the FFQ and evaluated as milligrams per day of calcium. For women, current
use of oral conjugated estrogen, patch, or cream at or within
a year of the baseline examination was determined from the
baseline Framingham Study questionnaire. Estrogen use
was defined as current users versus noncurrent users (including both former and never users), based on prior work
showing effects only for current estrogen use in elderly
women.(35) Too few of the elderly women were current
users to adjust further for dose of estrogen use.
Statistical analyses
Baseline characteristics were compared using Student’s
t-tests or ␹2 tests as appropriate. We evaluated each BMD
site separately, using linear regression to examine the relation of change in BMD with percent protein intake with
simultaneous adjustment for potential confounding variables. Because total energy intake is correlated with most
nutrients, we adjusted for total energy intake to assess the
independent effect contributed by protein. By doing so,
possible differences in intake caused by body size or activity levels also are taken into account.(28,36) The model
evaluated BMD change over the 4 years as a function of
percent protein intake, total energy intake, age, sex, and the
major known risk factors for bone loss: weight, weight
change, height, smoking, and alcohol use. We further adjusted for the potential additional effects of physical activity, calcium intake, and for women, current estrogen use
(yes/no).
We examined the effect of protein intake in the BMD
models using several well-established statistical methods(28):
(1) the residual method, regressing protein grams of intake
on total energy intake and used these residuals in the BMD
analyses; (2) the nutrient density model, adjusting grams of
protein intake for total energy intake in standard multivariate model; (3) the energy decomposition model, adjusting
protein intake for the total nonprotein energy intake; and,
finally, (4) protein not adjusted for total energy. Because all
these methods provided very similar results, only the results
from the analyses using percent protein adjusted for total
energy will be presented.
We evaluated percent protein as a continuous variable
and as quartiles of intake, to evaluate the possibility of a
nonlinear relation. The quartile analysis presents the protein
and BMD relation across the intake ranges seen in a population setting to examine the effect of possible levels of
protein intake on BMD loss, because it is unclear what
DIETARY PROTEIN AND BONE LOSS
2507
TABLE 1. COMPARISON OF BASELINE CHARACTERISTICS IN FRAMINGHAM COHORT MEMBERS WITH DIETARY FFQ DATA
WHO ATTENDED BOTH BASELINEa AND FOLLOW-UP EXAMINATIONS TO THOSE MEMBERS WHO ONLY
ATTENDED BASELINE EXAMINATION
Characteristics
Age (years)b
(range)
Weight (kg)b
(range)
Percent female
Protein intake (g/day)b
% Protein per day
Animal protein
Protein (g/kg per day)
Energy intake
BMD (g/cm2)b
Radial shaft
Femoral neck
Trochanter
Ward’s area
Lumbar spine
Percent current smoker
Former smoker
Never smoker
Alcohol intake (oz/week)b
Calcium intake (mg/day)b
Women only
Current estrogen use
Ever used estrogen
a
b
Attended
both exams
(n ⫽ 615)
Attended only
baseline exam
(n ⫽ 240)
74.5 ⫾ 4.4
(68–91)
70.6 ⫾ 14.7
(41–148)
64%
68.5 ⫾ 23.6
15.9 ⫾ 3.4
45.7 ⫾ 19.1
1.00 ⫾ 0.39
1736 ⫾ 572
77.2 ⫾ 5.3
(67–93)
68.4 ⫾ 13.2
(40–122)
55%
66.8 ⫾ 24.4
15.6 ⫾ 3.35
44.85 ⫾ 19.1
1.00 ⫾ 0.42
1732 ⫾ 620
0.001
0.594 ⫾ 0.134
0.789 ⫾ 0.144
0.709 ⫾ 0.161
0.606 ⫾ 0.149
1.169 ⫾ 0.231
10%
47%
43%
2.5 ⫾ 3.72
810 ⫾ 437.0
0.585 ⫾ 0.136
0.763 ⫾ 0.154
0.690 ⫾ 0.178
0.587 ⫾ 0.173
1.167 ⫾ 0.280
13%
47%
40%
2.3 ⫾ 3.92
783 ⫾ 417.0
0.35
0.03
0.14
0.16
0.91
0.36
0.61
0.40
7.0%
38.7%
3.0%
33.8%
0.09
0.31
p Value
0.04
0.01
0.41
0.27
0.56
0.88
0.96
Number for attendees based on having either femoral or radial shaft BMD scan at both exams.
Mean at baseline ⫾ SD.
intake levels may cause concern in the elderly. Men and
women had similar distributions of overall protein intake
and animal protein intake, as well as similar relations between protein intake and BMD change. There were no sex
by protein interactions in models for any bone site. Thus,
analyses for men and women were combined. For the quartile analyses, adjusted mean BMDs are presented (least
squares means ⫾ SE), for levels of protein intake resulting
from the analysis of covariance (ANCOVA). All analyses
used the SAS statistical analysis package (release 6.12; SAS
Institute Inc., Cary, NC, U.S.A.). Models of the absolute
change in BMD from baseline examination to follow-up
examination showed similar results as the percent change
analyses, and, thus, the absolute change analyses are not
presented here. First, we evaluated the impact of dietary
protein intake on bone loss. Second, we examined the
influence of animal protein versus nonanimal protein intakes.
RESULTS
For the 615 subjects (391 women and 224 men) with
longitudinal data, the mean age (⫾SD) at baseline was 75
years (⫾4.4 years) with a range from 68 to 91 years. Cohort
members without follow-up data (the majority of whom
died during the follow-up period) tended to be older, of
lower weight, and male, with baseline BMDs slightly lower
than those followed (e.g., femoral neck BMD, 0.763 vs.
0.789 and p ⫽ 0.02); however, their baseline protein intakes
did not differ from cohort members with 4-year follow-up
(Table 1). Participants and nonparticipants had similar distributions for smoking, alcohol intake, and calcium intake.
Table 2 shows the distributions for mean protein intake as
well as percent of energy from protein by sex. Mean protein
intake for the participants was 68 ⫾ 23.6 g/day (SD) with a
range from 17 to 152 g/day. Protein comprised 16%
(⫾3.4%; range, 7–27%) of total energy intake. Percent of
energy from animal protein was 10% (⫾3.5%).
Baseline BMD values at the femoral neck, trochanter,
Ward’s area, radial shaft, and lumbar spine were not significantly associated with protein intake, with p values ranging from 0.30 (trochanter and radius) to 0.54 (Ward’s area).
Mean 4-year BMD losses have been reported,(23) and for
women, ranged from ⫺4.84% (radial shaft) to ⫺3.42%
(trochanter), while losses for men ranged from ⫺3.59%
(radial shaft) to ⫺0.17% (trochanter).
Overall dietary protein intake: Lower percent protein
intake was significantly related to greater BMD loss at all
femur and spine sites (all p ⬍ 0.02), but not at the radial
2508
HANNAN ET AL.
TABLE 2. DISTRIBUTIONS
Protein intake
Protein intake (g/day)
Mean (⫾SD)
Quartile 1 low
2
3
4 high
Protein intake (g/kg per day)
Mean (⫾SD)
Quartile 1 low
2
3
4 high
Percent energy from protein (of
total caloric intake per day)
Mean % (⫾SD)
Quartile 1 low
2
3
4 high
Animal protein intake (g/day)
Mean (⫾SD)
Quartile 1 low
2
3
4 high
Percent animal protein intake (of
total caloric intake per day)
Mean % (⫾SD)
Quartile 1 low
2
3
4 high
OF
TYPES
OF
DIETARY PROTEIN INTAKE
BY
SEX
Men (n ⫽ 224)
Women (n ⫽ 392)
68 ⫾ 23.6
17–51
52–67
68–83
84–152
69.3 ⫾ 23.9
21–53
54–69
70–83
84–146
68.0 ⫾ 23.5
16–50
51–65
66–81
82–153
1.00 ⫾ 0.39
0.21–0.71
0.72–0.96
0.97–1.23
1.24–2.78
0.89 ⫾ 0.33
0.27–0.63
0.64–0.87
0.88–1.05
1.06–1.89
1.07 ⫾ 0.40
0.21–0.77
0.78–0.99
1.00–1.30
1.31–2.78
15.9 ⫾ 3.3
7.3–13.5
13.7–15.7
15.8–17.8
17.9–27.4
15.0 ⫾ 3.3
7.3–12.9
13.0–15.0
15.1–17.1
17.2–26.9
16.3 ⫾ 3.4
8.1–14.2
14.3–16.1
16.2–18.3
18.4–27.4
45.7 ⫾ 19.1
4–32
33–43
44–57
58–132
45.5 ⫾ 18.8
6–33
34–44
45–57
58–111
45.8 ⫾ 19.3
4–32
33–43
44–57
58–132
10.6 ⫾ 3.5
1.9–8.2
8.3–10.3
10.4–12.5
12.6–23.4
9.9 ⫾ 3.3
2.1–7.6
7.7–9.7
9.8–11.7
11.8–23.4
11.0 ⫾ 3.5
1.9–8.5
8.6–10.9
11.0–13.0
13.1–22.8
Total
shaft (p ⫽ 0.26). After adjusting for the major risk factors
for osteopenia, lower percent protein intake remained significantly related to greater BMD loss at the femoral neck
(␤ ⫽ 0.203; p ⫽ 0.02), Ward’s area (␤ ⫽ 0.266; p ⫽ 0.04),
and spine (␤ ⫽ 0.281; p ⫽ 0.02) with regression coefficients
at all three sites comparable with 10 lb of weight or a
smoking effect, both well-established risk factors for osteopenia. The p values for all the overall global ANCOVA
models ranged from 0.0001 for the femoral neck and lumbar
spine to 0.0628 for the radial shaft. Further adjustment for
physical activity, calcium intake, and for women, current
estrogen use, did not alter the effect of protein intake on
change in BMD.
When quartiles of percent protein intake were evaluated,
the lowest protein quartile showed the greatest bone loss.
(Fig. 1). Similar results were seen at the other femur sites
and the lumbar spine, with a similar trend observed for the
radial shaft. When the relation between BMD loss and
protein was adjusted for the major known risk factors for
bone loss—weight, weight change, height, age, sex, smoking, and alcohol use—the lowest quartile of protein continued to have the greatest BMD loss (Table 3). The highest
quartile, with protein intakes of 1.24 –2.78 g/kg. per day,
showed the least BMD loss across follow-up. Again, further
adjustment for physical activity, calcium intake, and for
women, current estrogen use, did not alter these results.
Animal protein intake: Similar to the overall protein
effect, lower percent of energy from animal protein also was
significantly related to bone loss at all femoral and spine
BMD sites (all p ⬍ 0.01) but not the radial shaft (p ⫽ 0.23).
Percent of energy from nonanimal protein did not contribute
to these BMD models (p value range, 0.79 – 0.98). When
these models were adjusted for the major risk factors for
osteopenia, the relation between BMD change and animal
protein intake remained statistically significant (femoral
neck, p ⫽ 0.03; lumbar spine, p ⫽ 0.02) in a manner
consistent with the overall protein effect described previously. Again, further adjustment of the model for physical
activity, calcium intake, and for women, current estrogen
use, did not alter the results.
When percent animal protein quartiles were evaluated,
the lowest quartile showed the greatest bone loss (Fig. 2),
similar to the overall protein intake results for quartiles,
although no relation was seen between radial shaft bone loss
and animal protein intake. Table 4 displays the least squares
mean results for all BMD sites and, again, these results
DIETARY PROTEIN AND BONE LOSS
FIG. 1. Mean percent bone loss over 4 years (⫾SE) at hip,
spine, and radius by quartiles of protein intake (Framingham
Osteoporosis Study).
parallel those seen in Table 3. None of the BMD sites
indicated a relation between BMD and nonanimal protein.
Figure 3 shows the lack of a relation between nonanimal
protein intakes (quartiles) and bone loss at the femoral neck.
DISCUSSION
The mean protein intake of 68 g/day and percent protein
intake of 16% for the study participants are similar to
recommendations for total protein and percent of energy
from protein set by the U.S. Government and also similar to
other reported values.(37,38) The current recommended daily
allowance (RDA) for protein intake is 0.8 mg/kg, and
roughly 32% of the Framingham Cohort have protein intake
below this RDA; however, 90% have an intake at least
two-thirds of the RDA (two-thirds of RDA is a common
cut-off used to estimate dietary adequacy, because the RDA
contains a margin of error to cover most healthy individuals). In our study, both lower protein intake and lower
animal protein intake were significantly related to greater
BMD loss at femur and spine BMD sites. Even after adjustment for the major risk factors for bone loss, including
weight, weight loss, and smoking, lower, but not higher,
protein intake remained significantly related to greater
BMD loss at the femur and spine with effects comparable
with 10 lb of weight or a smoking effect, both wellestablished risk factors for osteopenia. Those elders with
higher protein intake had reduced bone loss, suggesting that
protein intake is important in maintaining bone or minimizing bone loss in elderly persons.
Contrary to expectations, elders with animal protein intake up to several-fold greater than the RDA also had the
least bone loss after controlling for known confounders.
Nonanimal sources of protein were not related to BMD.
These results suggest that typical population intakes of
animal protein, within the range commonly consumed, do
not result in bone loss. Rather animal protein intake appears
important in maintaining bone or minimizing bone loss in
elderly persons.
2509
A number of studies(4,39 – 41) reported that a doubling of
protein intake increases urinary calcium loss by 50%. Parfitt
also noted that the acid load from dietary protein is partially
buffered by skeletal bone loss, accounting for a portion of
age-related bone loss.(42) Allen reported that urinary calcium loss is correlated directly with dietary intake of protein
and that high calcium diets do not prevent the negative
calcium balance and bone loss induced by a diet high in
protein,(43) although it is unclear what levels of protein
intake would be considered high for this population. The
influence of dietary protein metabolically may be not as
great in the elderly as one would assume, based on estimated intake, because additional age-related changes in
renal function and intestinal absorption influence calcium
imbalance.(44) Although these studies examined short-term
calcium loss, several cross-sectional studies of forearm
BMD and protein intake reported no association between
dietary protein and BMD.(14,45)
Our findings are in agreement with two papers reporting
better bone health in women with greater protein intakes.
Freudenheim et al.(46) showed that high protein intake protected against low radius BMD in older women. Munger et
al. found an increased risk of hip fracture in elderly women
consuming the lowest amounts of protein in the Iowa Women’s Health Study.(47) Further, the Munger study reported
that higher intakes of animal sources of dietary protein were
associated with a 70% reduction in hip fracture, even after
controlling for major confounding variables.
A number of studies support the adverse role of low
protein intake on bone metabolism. Hirota et al.(48) reported
that low protein intake in young Asian women was a risk
factor for low forearm BMD in their study of 161 women
ages 19 –25 years. Similarly, Geinoz et al.,(49) in a study of
74 hospitalized geriatric patients (mean age, 82 years),
showed that low dietary protein intake was associated with
low femoral BMD. Additional studies of protein supplementation in elderly women posthip fracture clearly showed
benefit in terms of BMD and muscle strength,(50,51) implying that protein insufficiency, particularly in the oldest of
old, contributes to osteoporosis. In addition, two studies
reported that hip fracture patients had diets particularly
deficient in protein and energy.(17,18) Kerstetter et al. show
that low protein intake may induce secondary hyperparathyroidism, perhaps inducing loss of bone.(52,53) Indeed,
there is potential for interventions with protein intakes.
Schurch et al. conducted a 6-month randomized clinical trial
of protein supplementation of 20 g/day in hip fracture
patients and showed 50% attenuation in femoral bone loss
after 1 year of follow-up.(54)
Orwoll et al. examined vertebral and radius BMDs and
reported that protein undernutrition, as assessed by serum
albumen measures, was associated with osteopenia, and that
low dietary protein intakes possibly influence bone metabolism.(55) They noted that although excess dietary protein
has been shown to cause negative calcium balance, this
occurs primarily with extremely high levels of protein, not
often seen in the elderly who are at the highest risk for
osteoporosis. Indeed, Chu et al., in a trial that doubled the
typical protein intake in elders, reported that rather than
causing negative calcium balance, the increased protein
2510
HANNAN ET AL.
TABLE 3. QUARTILES
OF
PERCENT PROTEIN INTAKE BY ADJUSTED LEAST SQUARES MEAN PERCENT BMD
CHANGE AT HIP, SPINE, AND RADIUS SITES
Site
Quartile 1
Quartile 2
Quartile 3
Quartile 4
Femoral neck
Trochanter
Ward’s area
Lumbar spine
Radius
⫺4.61 ⫾ 0.70*
⫺8.00 ⫾ 0.84
⫺7.05 ⫾ 1.0†
⫺3.72 ⫾ 0.97†
⫺4.21 ⫾ 0.71
⫺3.68 ⫾ 0.71‡
⫺7.70 ⫾ 0.86
⫺6.14 ⫾ 1.0
⫺1.76 ⫾ 1.0
⫺5.33 ⫾ 0.73
⫺3.05 ⫾ 0.73
⫺7.28 ⫾ 0.89
⫺5.41 ⫾ 1.0
⫺2.40 ⫾ 1.0
⫺3.82 ⫾ 0.75
⫺2.32 ⫾ 0.74
⫺6.65 ⫾ 0.90
⫺4.39 ⫾ 1.1
⫺1.11 ⫾ 1.1
⫺4.31 ⫾ 0.76
Model adjusted for total energy intake, age, sex, weight, weight change, height, alcohol intake and smoking (current or former).
Each quartile compared with quartile 4: *0.001 ⬍ p ⬍ 0.0001; † 0.05 ⬍ p ⬍ 0.01; ‡ 0.10 ⬍ p ⬍ 0.05.
FIG. 2. Mean percent bone loss over 4 years (⫾SE) at hip,
spine, and radius by quartiles of animal protein intake
(Framingham Osteoporosis Study).
intakes improved the calcium balance in the majority of
their subjects.(56) These studies raise the possibility that low
protein intakes, particularly in the elderly, may adversely
affect bone mineral metabolism.
These studies, when considered together, suggest that
adequate protein nutriture is required for bone health. Metabolic studies showing that high intakes of protein may have
an adverse consequence on calcium balance have all been of
short-term duration. It remains unclear what the long-term
influences of protein intake and its possible acid load have
on bone. It may well be that only extreme excess protein
intake or deficient protein intake may be deleterious yet
uncommon problems in human populations. Our results
suggest that within the normal variation in dietary protein of
this population of elders low protein intake is associated
with BMD loss, while higher (normal) protein intake is
associated with reduced bone loss or with maintenance of
BMD.
This study has several unique strengths. First, the Framingham cohort is population based, rather than a study of
volunteers or those who already have disease. Second, the
longitudinal design provides important information about
factors responsible for bone loss in old age. Third, unlike
most studies of osteoporosis, it includes a large number of
men as well as women who have longitudinal BMD measurements. Fourth, this study has comprehensive dietary
assessments that have shown to estimate usual nutrient
intake. Finally, the finding of a consistent protein effect at
multiple bone sites after control of potential confounders
strongly suggests that adequate protein intake is important
to skeletal health of aged persons.
Several limitations also should be noted. The data do not
take into account possible longer-term effects of certain risk
factors (e.g., cumulative effect of smoking or medications).
Second, different technology assessed femoral BMD at
baseline and follow-up, although data were “standardized.”
Finally, we cannot derive precise protein intake values
representing the lowest threshold, because we used an FFQ
that typically best ranks individuals’ dietary intakes.
As the population ages, osteoporosis will escalate in
importance as a major public health problem. This study
suggests an important, potentially modifiable factor that
could have major implications for the diets of millions of
men and women as well as affect their risk of osteoporosis.
Our results indicate that low protein intake is associated
with BMD loss. Campbell et al.(57) make a compelling
argument that the protein RDA for older persons in the
United States, established from extrapolations from healthy
young men, is too low. Based on several protein requirement studies of elderly subjects, they recommend a safe
protein intake to range from 1.0 to 1.25 g/kg per day. These
values would correspond roughly to the third quartile in our
analyses. A metabolic balance study(58) and a nutritional
status survey of elders over the age of 65 years(59) also
provide strong support that elders have a higher protein
requirement than that currently recommended. Even so, the
U.S. Department of Agriculture (USDA) data on nutrition
report that 30% of all adult women are below the RDA for
protein and over 25% of the Framingham cohort have
protein intake below the current RDA of 0.6 g/kg. If the
protein intake sufficiency threshold is even higher for elders
than the current RDA, low protein intake may place many
elders at risk for bone loss.
This population-based study suggests that dietary protein
intake is an important component of bone health in elders,
showing an effect for both women and men with age-related
bone loss, even after controlling for major known confounders. Further, this study indicates that high levels of dietary
protein intake, within the range commonly consumed, do
not result in bone loss in elders. Ensuring adequate dietary
protein intake is an important component of bone health in
elders.
DIETARY PROTEIN AND BONE LOSS
TABLE 4. QUARTILES
OF
2511
PERCENT ANIMAL PROTEIN INTAKE AND ADJUSTED LEAST SQUARES MEAN PERCENT BMD CHANGE
AT HIP, SPINE, AND RADIUS SITES
Site
Quartile 1
Quartile 2
Quartile 3
Quartile 4
Femoral neck
Trochanter
Ward’s area
Lumbar spine
Radius
⫺3.95 ⫾ 0.69*
⫺2.57 ⫾ 0.86
⫺4.02 ⫾ 1.0†
⫺3.79 ⫾ 0.99†
⫺4.60 ⫾ 0.71
⫺3.83 ⫾ 0.72*
⫺3.25 ⫾ 0.89
⫺3.90 ⫾ 1.1
⫺2.55 ⫾ 1.0
⫺4.35 ⫾ 0.73
⫺3.02 ⫾ 0.73
⫺1.75 ⫾ 0.91
⫺2.03 ⫾ 1.1
⫺1.24 ⫾ 1.0
⫺4.39 ⫾ 0.77
⫺2.15 ⫾ 0.73
⫺1.95 ⫾ 0.92
⫺1.97 ⫾ 1.1
⫺1.65 ⫾ 1.1
⫺4.52 ⫾ 0.76
Model adjusted for total energy intake, nonanimal protein intake, age, sex, weight, weight change, height, alcohol intake and smoking
(current or former).
Each quartile compared with quartile 4: *0.05 ⬍ p ⬍ 0.01; † 0.10 ⬍ p ⬍ 0.05.
8.
9.
10.
11.
12.
FIG. 3. Mean percent femoral neck bone loss over 4 years
(⫾SE) by quartiles of nonanimal protein intake (Framingham Osteoporosis Study).
13.
14.
ACKNOWLEDGMENTS
We are grateful to the Framingham cohort participants
and staff, and we also thank the densitometer technicians
Mimi Brodsky, Mary Hogan, and Cherlyn Mercier. This
work was supported in part by a New Investigator Award
from the Arthritis Foundation, National Institutes of Health
(NIH) grants RO1-AR/AG41398 and AR20613, and U.S.
Department of Agriculture (USDA) Agricultural Research
Service (ARS) contract number 53-3K06-01.
15.
16.
17.
18.
REFERENCES
1. Nevitt M 1994 Epidemiology of osteoporosis. Rheum Dis Clin
North Am 20:535–559.
2. Melton LJ III 1995 How many women have osteoporosis now?
J Bone Miner Res 10:175–177.
3. Kanis JA 1994 Assessment of fracture risk and its application
to screening for postmenopausal osteoporosis: Synopsis of a
WHO report. Osteoporos Int 4:368 –381.
4. Heaney RP 1993 Nutritional factors in osteoporosis. Annu Rev
Nutr 13:287–316.
5. Sherman H 1920 Calcium requirements of maintenance in
man. J Biol Chem 44:21–27.
6. Pullman R, Alving AS, Denn RJ, Landowne M 1954 The
influence of dietary protein on specific renal functions in
normal man. J Lab Clin Med 44:320 –332.
7. Anand CR, Linkswiler HM 1974 Effect of protein intake on
19.
20.
21.
22.
23.
calcium balance of young men given 500 mg calcium daily. J
Nutr 104:695–700.
Johnson NE, Alcantara EN, Linkswiler H 1970 Effect of level
of protein intake on urinary and fecal calcium and calcium
retention of young adult males. J Nutr 100:1425–1430.
Margen S, Chu JY, Kaufmann NA, Calloway DH 1974 Studies
in calcium metabolism. I. The calciuretic effect of dietary
protein. Am J Clin Nutr 27:584 –589.
Walker RM, Linkswiler HM 1972 Calcium retention in the adult
human male as affected by protein intake. J Nutr 102:1297–1302.
Sebastian A, Harris ST, Ottaway JH, Todd KM, Morris RC Jr
1994 Improved mineral balance and skeletal metabolism in
postmenopausal women treated with potassium bicarbonate.
N Engl J Med 330:1776 –1781.
Tylavsky FA, Anderson JJ 1988 Dietary factors in bone health
of elderly lacto-ovovegetarian and omnivorous women. Am J
Clin Nutr 48(Suppl 3):842– 849.
Abelow BJ, Holford TR, Insogna KL 1992 Cross-cultural
association between dietary animal protein and hip fracture: A
hypothesis. Calcif Tissue Int 50:14 –18.
Hunt IF, Murphy NJ, Henderson C, Clark VA, Jacobs RM,
Johnston PK, Coulson AH 1989 Bone mineral content in
postmenopausal women: Comparison of omnivores and vegetarians. Am J Clin Nutr 50:517–523.
Patterson B, Cornell CN, Carbone B, Levine B, Chapman D
1992 Protein depletion and metabolic stress in elderly patients
who have a fracture of the hip. Am J Bone Joint Surg 74:251–
260.
Bonjour J, Schurch MA, Rizzoli R 1996 Nutritional aspect of
hip fractures. Bone 18(Suppl 3):139S–144S.
Jensen JE, Jensen TG, Smith TK, Johnston DA, Dudrick SJ
1982 Nutrition in orthopaedic surgery. J Bone Joint Surg Am
64:1263–1272.
Bastow MD, Rawlings J, Allison SP 1983 Undernutrition, hypothermia, and injury in elderly women with fractured femur: An
injury response to altered metabolism? Lancet 1:143–146.
Dawber T, Meadors, GF, Moore, FE JR 1951 Epidemiological
approaches to heart disease: The Framingham Study. Am J
Public Health 41:279 –286.
Dawber T 1980 The Framingham Study. Harvard University
Press, Cambridge, MA, U.S.A.
Shurtleff D 1970 The Framingham Study: An epidemiological
investigation of cardiovascular disease. In: Kannel W, Gordon,
T (eds.) U.S. Government Printing Office, Washington D.C.,
U.S.A.
Hannan MT, Felson DT, Anderson JJ 1992 Bone mineral
density in elderly men and women: Results from the Framingham Osteoporosis Study. J Bone Miner Res 7:547–553.
Hannan M, Felson DT, Dawson-Hughes B, Tucker KL,
Cupples LA, Wilson PWF, Kiel DP 2000 Risk factors for
longitudinal bone loss in elderly men and women: The Framingham Osteoporosis Study. J Bone Miner Res 15:710 –720.
2512
24. Kiel DP, Mercier CA, Dawson-Hughes B, Cali C, Hannan MT,
Anderson JJ 1995 The effects of analytic software and scan
analysis technique on the comparison of dual X-ray absorptiometry with dual photon absorptiometry of the hip in the
elderly. J Bone Miner Res 10:1130 –1136.
25. Rimm E, Giovannucci EL, Stampfer MJ, Colditz GA, Litin
LB, Willett WC 1992 Reproducibility and validity of an expanded self-administered semi-quantitative food frequency
questionnaire among male health professionals. Am J Epidemiol 135:1114 –1126.
26. Willett WC, Sampson L, Stampfer MJ, Rosner B, Bain C,
Witschi J, Hennekens CH, Speizer FE 1985 Reproducibility
and validity of a semiquantitative food frequency questionnaire. Am J Epidemiol 122:51– 65.
27. Jacques PF, Sulsky SI, Sadowski JA, Phillips JC, Rush D,
Willett WC 1993 Comparison of micronutrient intake measured by a dietary questionnaire and biochemical indicators of
micronutrient status. Am J Clin Nutr 57:182–189.
28. Willett W 1990 Nutritional Epidemiology Monographs in Epidemiology & Biostatistics, vol. 13. Oxford University Press,
New York, NY, U.S.A.
29. Kiel DP, Felson DT, Hannan MT, Anderson JJ, Wilson PW
1990 Caffeine and the risk of hip fracture: The Framingham
Study. Am J Epidemiol 132:675– 684.
30. Cummings SR, Nevitt MC, Browner WS, Stone K, Fox KM,
Ensrud KE, Cauley J, Black D, Vogt TM 1995 Risk factors for
hip fracture in white women. Study of Osteoporotic Fractures
Research Group. N Engl J Med 332:767–773.
31. Felson DT, Kiel DP, Anderson JJ, Kannel WB 1988 Alcohol
consumption and hip fractures: The Framingham Study. Am J
Epidemiol 128:1102–1110.
32. Felson DT, Zhang Y, Hannan MT, Kannel WB, Kiel DP 1995
Alcohol intake and bone mineral density in elderly men and
women. The Framingham Study. Am J Epidemiol 142:485– 492.
33. Hannan MT, Felson DT, Anderson JJ, Naimark A 1993 Habitual physical activity is not associated with knee osteoarthritis: The Framingham Study. J Rheumatol 20:704 –709.
34. Kannel WB, Sorlie P 1979 Some health benefits of physical
activity. The Framingham Study. Arch Intern Med 139:857–
861.
35. Felson DT, Zhang Y, Hannan MT, Kiel DP, Wilson PW,
Anderson JJ 1993 The effect of postmenopausal estrogen
therapy on bone density in elderly women. N Engl J Med
329:1141–1146.
36. Willett W, Stampfer MJ 1986 Total energy intake: Implications for epidemiologic analyses. Am J Epidemiol 124:17–27.
37. Smit E, Nieto FJ, Crespo CJ, Mitchell P 1999 Estimates of
animal and plant protein intake in U.S. adults: Results from the
Third National Health and Nutrition Examination Survey,
1988 –1991. J Am Diet Assoc 99:813– 820.
38. McDowell M, Briefel RR, Alaimo K, Bischof AM, Caughman
CR, Carroll MD, Loria CM, Johnson CL 1994 Energy and
macronutrient intakes of persons ages 2 months and over in the
United States: Third National Health and Nutrition Examination Survey, phase I, 1988 –1991. Advance data from Vital and
Health Statistics 255. National Center for Health Statistics,
Hyattsville, MD, U.S.A.
39. Linkswiler HM, Zemel MB, Hegsted M, Schuette S 1981
Protein-induced hypercalciuria. Fed Proc 40:2429 –2433.
40. Lutz J, Linkswiler HM 1981 Calcium metabolism in postmenopausal and osteoporotic women consuming two levels of
dietary protein. Am J Clin Nutr 34:2178 –2186.
41. Angus RM, Sambrook PN, Pocock NA, Eisman JA 1988
Dietary intake and bone mineral density. Bone Miner 4:265–
277.
42. Parfitt AM 1983 Dietary risk factors for age-related bone loss
and fractures. Lancet 2:1181–1185.
43. Allen LH, Oddoye EA, Margen S 1979 Protein-induced hypercalciuria: A longer term study. Am J Clin Nutr 32:741–749.
HANNAN ET AL.
44. Massey LK 1993 Dietary factors influencing calcium and bone
metabolism: Introduction. J Nutr 123:1609 –1610.
45. Sowers MR, Wallace RB, Lemke JH 1985 Correlates of midradius bone density among postmenopausal women: A community study. Am J Clin Nutr 41:1045–1053.
46. Freudenheim JL, Johnson NE, Smith EL 1986 Relationships
between usual nutrient intake and bone-mineral content of
women 35– 65 years of age: Longitudinal and cross-sectional
analysis. Am J Clin Nutr 44:863– 876.
47. Munger Cerhan JR, Chiu BC 1999 Prospective study of dietary
protein intake and risk of hip fracture in postmenopausal
women. Am J Clin Nutr 69:147–152.
48. Hirota T, Nara M, Ohguri M, Manago E, Hirota K 1992 Effect
of diet and lifestyle on bone mass in Asian young women.
Am J Clin Nutr 55:1168 –1173.
49. Geinoz G, Rapin CH, Rizzoli R, Kraemer R, Buchs B, Slosman D, Michel JP, Bonjour JP 1993 Relationship between
bone mineral density and dietary intakes in the elderly. Osteoporos Int 3:242–248.
50. Delmi M, Rapin CH, Bengoa JM, Delmas PD, Vasey H,
Bonjour JP 1990 Dietary supplementation in elderly patients
with fractured neck of the femur. Lancet 335:1013–1016.
51. Tkatch L, Rapin CH, Rizzoli R, Slosman D, Nydegger V,
Vasey H, Bonjour JP 1992 Benefits of oral protein supplementation in elderly patients with fracture of the proximal femur.
J Am Coll Nutr 11:519 –525.
52. Kerstetter JE CD, Mitnick ME, Ellison AF, Gay LF, Liskov T,
Carpenter TO, Insogna KL 1997 Increased circulating concentrations of parathyroid hormone in healthy, young women
consuming a protein-restricted diet. Am J Clin Nutr 66:1188 –
1196.
53. Kerstetter JE, O’Brien KO, Insogna KL 1998 Dietary protein
affects intestinal calcium absorption. Am J Clin Nutr 68:859 –
865.
54. Schurch MA, Rizzoli R, Slosman D, Vadas L, Vergnaud P,
Bonjour JP 1998 Protein supplements increase serum insulinlike growth factor I levels and attenuate proximal femur bone
loss in patients with recent hip fracture. Ann Intern Med
128:801– 809.
55. Orwoll ES, Weigel RM, Oviatt SK, Meier DE, McClung MR
1987 Serum protein concentrations and bone mineral content
in aging normal men. Am J Clin Nutr 46:614 – 621.
56. Chu JY, Margen S, Costa FM 1975 Studies in calcium metabolism. II. Effects of low calcium and variable protein intake on
human calcium metabolism. Am J Clin Nutr 28:1028 –1035.
57. Campbell WW, Crim MC, Dallal GE, Young VR, Evans WJ
1994 Increased protein requirements in elderly people: New
data and retrospective reassessments. Am J Clin Nutr 60:501–
509.
58. Gersovitz M, Motil K, Munro HN, Scrimshaw NS, Young VR
1982 Human protein requirements: Assessment of the adequacy of the current recommended dietary allowance for dietary protein in elderly men and women. Am J Clin Nutr
35:6 –14.
59. Hartz S, Russell RM, Rosenberg IH 1992 Nutrition in the
elderly: The Boston nutritional status survey. Smith-Gordon
and Company, Ltd., London, U.K.
Address reprint requests to:
Marian T. Hannan
Hebrew Rehabilitation Center for Aged Research
and Training Institute
1200 Center Street
Boston, MA 02131-1097, U.S.A.
Received in original form March 3, 2000; in revised form July 14,
2000; accepted August 16, 2000.